INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

Diphtheria toxin (DT) production by Corynebacterium diphtheriae is mediated by phage conversion (6, 7), and the structural gene for DT, tox, is present in the genomes of corynebacteriophages such as phage  (1, 37). DT production is influenced by the amount of iron in the growth medium (11, 16), is maximal under conditions of iron starvation, and is significantly decreased under high-iron conditions (5, 13). The DT repressor (DtxR) is a negative regulator that controls the expression of DT and the high-affinity iron uptake system of C. diphtheriae, in addition to other C. diphtheriae gene products (10, 15, 24, 25, 28, 31-33, 35).

The formation of DtxR homodimers is mediated by protein-protein interactions between the monomers, and binding of divalent cations is required for the activation of repressor activity (2, 25, 27, 29, 33, 36, 39). Numerous genetic studies have focused on defining the metal binding domain(s) of DtxR and the mechanism(s) by which divalent cations activate the repressor (3, 34, 39). Saturation site-directed mutagenesis demonstrated that C102 is essential for the activation of DtxR (34), and subsequent random and site-directed mutagenesis studies identified other residues that are important for function of the DNA-binding motif in domain 1 and the two metal binding sites in domain 2 (3, 39).

The crystal structure of dimeric DtxR was initially solved at a 2.8-Å resolution in complex with six different divalent transition metals (19). Each subunit was shown to possess an amino-terminal domain that includes a DNA-binding helix-turn-helix motif; an interface domain containing two distinct metal binding sites, which lie 10 Å apart; and a third, flexible carboxyl-terminal domain, which was later shown to be an SH3-like domain that did not appear to be involved in either DNA or metal binding by DtxR (20). High-resolution structures of DtxR complexed with cobalt (2.0 Å) and manganese (2.2 Å) revealed that metal binding site 1 was well occupied in both structures and demonstrated that a sulfate ion served as the fourth ligand for the divalent cation at metal binding site 1 (20). The sulfate anion was also shown to participate in an extensive network of hydrogen bonds with other ligands in DtxR (20). DtxR was crystallized with cobalt at 100 K with a 1.85-Å resolution, providing the highest resolution to date, and at room temperature with zinc at a 2.4-Å resolution, revealing no significant differences between the two structures but providing the most accurate view of the anion/cation binding site (17). In the 1.85-Å cobalt-sulfate-DtxR structure, metal binding site 1 was coordinated tetrahedrally by the N² of H79, the O¹ of E83, the N¹ of H98, and a sulfate ion. In addition, the N of R80, the O of S126, and the O¹ of N130 were involved in the coordination of the sulfate ion (Fig. 1). The zinc-DtxR crystal was prepared in the absence of sulfate, yet a tetrahedral molecule consistent with a phosphate anion was present in the structure at the anion binding site (17). These results suggested that under physiological conditions phosphate may function as a co-corepressor for DtxR.

View larger version (28K): [in this window] [in a new window]   FIG. 1.   Close-up view of the anion/cation binding site in the cobalt-DtxR structure determined at a 1.85-Å resolution (17) showing the coordination of the Co2+ and sulfate ions in addition to the interaction between E20 in domain 1 and R80 in the anion/cation binding site of domain 2. Hydrogen bonds are depicted as dashed lines. The helices depicted in green belong to the dimerization domain, and the helix shown in violet represents the DNA binding domain of DtxR. One of the sulfate-coordinating ligands, N130, was omitted (17) to show more clearly the E20-R80 interaction.

Metal binding site 1 was fully occupied in all six of the DtxR-cation complexes, as described above (19, 20). In contrast, metal binding site 2 was highly occupied in wild-type DtxR only when it was in complex with CdCl₂, and the metal-binding ligands included the carbonyl group of C102, the O¹ atom of E105, the N² atom of H106, and a solvent molecule. The side chain of C102 did not interact with the metal at site 2 but rather reacted with the imidazole ring of H98, one of the direct ligands of metal binding site 1 (19). In these crystals, however, the SH group of C102 appeared to be oxidized (19, 20) or covalently modified (17), which may possibly have altered its capacity to interact with the divalent metal ion at site 2. The structure of a C102D substitution variant of DtxR that retained biological activity was also determined when it was in complex with Ni²⁺ (3). Both sites 1 and 2 exhibited high occupancy by metal ions in DtxR-C102D, and the metal-binding ligands at site 2 were reported to include the side chains of D102 and M10 and a solvent molecule in addition to the carbonyl oxygen of D102 and the side chains of E105 and H106.

The contributions of the two metal binding sites to the activity of DtxR have been controversial. Variants with alanine substitutions for each of the metal-binding ligands at site 1 and site 2 in DtxR were constructed by site-directed mutagenesis, and only the substitutions at site 2 ligands were found to cause dramatic inactivation of DtxR activity in an Escherichia coli reporter system (3). More recently, the structure of metal-activated DtxR-C102D in complex with an oligonucleotide containing the tox operator sequence was reported at a 3-Å resolution (40). Two dimers of the activated repressor bind to the operator on opposite faces of the DNA, and the main-chain oxygen of residue L4 forms a hydrogen bond with a solvent molecule which is itself a ligand of the metal ion at site 2. Based on these findings, it was proposed that metal binding at site 2 is primarily responsible for activating the repressor activity of DtxR, and the conformational change resulting in activation was proposed to involve a caliper-like rigid body rotation of the DtxR monomers with respect to one another, resulting in decreased distance between their DNA-binding motifs, as well as a helix-to-coil transition at the extreme amino terminus of each DtxR monomer (3, 40).

Both apo-DtxR and holo-DtxR can occur in either of two crystal forms, and structures of form 1 and form 2 crystals of apo-DtxR and holo-DtxR in the presence of zinc were recently determined and compared at resolutions of 2.2 to 2.4 Å (18). The N-terminal DNA binding domain and the last 20 amino acids of the dimerization domain of each subunit were shown to exhibit significant movement with respect to the immobile dimer core as a consequence of binding divalent transition metals. Activation of DtxR, therefore, involves a change in the tertiary structure of DtxR (18) rather than a change in the quaternary structure as proposed by other investigators (3, 23). In addition, both metal binding sites are occupied in activated forms of DtxR; as determined at resolutions or 2.2 Å or less, an anion is present at site 1 in structures of DtxR in complex with metal ions; the R80 residue of binding site 1 interacts with E20 of the DNA binding domain of DtxR (20) (Fig. 1); and the anion-binding ligands Arg80, Ser126, and Asn130 are all conserved in the reported homologs of DtxR (4, 8, 14, 30). Taken together, these findings suggest that anion/cation binding site 1 is an important structural feature of DtxR, and it remains likely that sites 1 and 2 in domain 2 are both important for DtxR function, notwithstanding the conclusions reported previously (3, 40).

In the present study, we used site-directed mutagenesis to construct variants of DtxR with alanine substituted for each of the anion-binding residues at site 1, and we compared the repressor activities of these variants with those of DtxR variants containing alanine substitutions for each of the metal-binding residues at site 1 or site 2 as controls. We also determined the effects of multiple alanine substitutions for these residues, investigated the importance of the interaction between R80 and E20 for DtxR repressor activity, and established the effects of the single or multiple alanine substitutions on the intracellular levels of the variant forms of DtxR.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Bacterial strains, plasmids, and media. The bacterial strains and plasmids used in this study are listed in Table 1. E. coli strains were cultured at 30°C in Luria-Bertani (LB) broth or on LB agar with ampicillin (100 µg/ml), chloramphenicol (30 µg/ml), or tetracycline (25 µg/ml) as needed. The ClpP protease-deficient E. coli strain SG22098 was constructed by Susan Gottesman at the National Institutes of Health and was kindly provided by Dorothy E. Pierson at the University of Colorado Health Sciences Center.

Construction of recombinant plasmids. (i) Site-specific mutagenesis. Oligonucleotides containing specific nucleotide substitutions, with respect to the coding sequence for wild-type dtxR, were purchased from Gibco BRL, Gaithersburg, Md., and are shown in Table 2. For single amino acid substitutions, site-specific mutagenesis was performed with the Muta-Gene in vitro mutagenesis kit (Bio-Rad, Hercules, Calif.) and single-stranded DNA generated from pSKIIdtxR in E. coli CJ236. To introduce multiple mutations, single-stranded DNA from clones known to have specific single mutations was subjected to one or more additional rounds of site-directed mutagenesis with different mutagenic oligonucleotides. Putative mutant clones were analyzed by DNA sequencing to confirm the presence of the predicted mutation(s).

(ii) T7 expression constructs. For expression in E. coli, wild-type or mutagenized variants of dtxR were transferred into the low-copy-number plasmid vector pWKS30 (38). The resulting constructs possessed various dtxR alleles under the transcriptional control of two T7 promoters. Such constructs express DtxR constitutively at low levels in E. coli in the absence of T7 RNA polymerase (unpublished results).

DNA sequencing and analysis. Double-stranded plasmid DNA for sequencing was isolated from E. coli DH5 clones, and for each clone the sequence of the segment of the dtxR gene that had been subjected to mutagenesis was determined by the dideoxy chain termination method of Sanger et al. (22). Dideoxy chain termination reactions were performed with T7 polymerase (Sequenase 2.0; United States Biochemicals, Cleveland, Ohio) by using primers MW-1 (5'[326]-CTCATAACGTGTTCCCAGC-3'[308]), MW-2 (5'[555]-AGCATCGAGGAGCTGTGTA-3'[537]), and MCS-2 (5'[158]-TTGTCGTTGTCGCCTCAGA-3'[176]), described previously (39). Reaction products were resolved on 8 M urea-6.6% polyacrylamide gels (21) and visualized by autoradiography.

DtxR antigen analysis. E. coli DH5(pCMZ100) and SG22098(pTXZ184) reporter strains expressing wild-type or variant forms of DtxR from pWKS30 constructs were grown overnight in LB broth with aeration at 30°C. Samples containing 10⁹ cells were centrifuged, and the pellets were resuspended in 125 µl of 1× sample buffer (21) and boiled for 5 min. Total cell samples were subjected to sodium dodecyl sulfate-12% polyacrylamide gel electrophoresis and Western blot analyses. A sample of polyclonal rabbit antiserum previously raised against a DtxR-MalE fusion protein (29) was adsorbed with an acetone powder prepared from E. coli DH5 (9) before being used to probe blots at a 1:10,000 dilution, followed by a secondary horseradish peroxidase-conjugated goat anti-rabbit antibody (Pierce, Rockford, Ill.). The immobilized immune complexes bound to wild-type or mutant forms of DtxR were detected with a chromogenic substrate for the enzyme activity (Renaissance kit; DuPont NEN, Boston, Mass.). The amount of each immunoreactive DtxR variant was determined by scanning the Western blots and analyzing the digitized data by using imaging software from the National Institutes of Health (NIH Image 1.55), and the expression level for each variant was expressed as a percentage relative to wild-type DtxR, taken as 100%. Control Western blots with samples containing serial dilutions of cells expressing wild-type DtxR demonstrated that the amount of immunoreactive DtxR detected was proportional to the number of cells applied and that the signals obtained for the DtxR variants were in the linear range of the assay.

Measurement of -galactosidase activity. Because the toxO/P-lacZ translational fusion in pCMZ100 or pTXZ184 is negatively regulated by DtxR, intracellular levels of -galactosidase activity decreased as DtxR repressor activity increased. -Galactosidase activity levels were determined as previously described by using o-nitrophenyl--D-galactopyranoside as a substrate (12, 21). Levels of -galactosidase activity were measured in Miller units.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

Effect of single site-directed mutations on DtxR activity. Single substitutions of alanine were made for the anion-binding residues of site 1 (R80A, S126A, and N130A), the metal-binding residues of site 1 (H79A, E83A, and H98A), and the metal-binding residues of site 2 (C102A, E105A, and H106A). Based on the crystallographic finding that residue R80 of domain 2 of DtxR interacts with E20 of domain 1 (17, 20), a single alanine substitution for E20 was also constructed. In addition, selected double- and triple-substitution variants of DtxR were made. The phenotypic effects of these substitutions on the production and activity of DtxR in E. coli were investigated by using a previously described reporter gene system that is responsive to regulation by DtxR and iron (25, 26).

Comparison of the effects of substitutions for residues R80 and E20. Recent crystallographic studies also demonstrated that R80 forms hydrogen bonds not only with the sulfate anion in site 1 but also with E20 in domain 1 of DtxR (Fig. 1). To determine whether the interaction between E20 and R80 is required for DtxR activity, the effects of alanine substitutions for E20 and R80 were tested both individually and in combination with alanine substitutions for other cation- or anion-binding residues in site 1 (Table 4). It was presumed that if an interaction between R80 and E20 is important for the function of the DtxR molecule, an E20A substitution might give the same results as an R80A substitution. This proved to be the case, as is indicated in Table 4. The effects of the single R80A and E20A substitutions on DtxR activity were indistinguishable and resulted in a dramatic decrease of repressor activity. It is important to note that random substitutions within the first domain of DtxR do not necessarily affect the activity of DtxR. In 1994, it was shown that several amino acid substitutions within domain 1, including T7I, R13C, E19K, T24I, T44I, and T67I, had little to no effect on DtxR activity (39).

Effect of multiple substitutions within DtxR metal binding site 1. Single amino acid substitutions of the metal-coordinating ligands of site 1 did not affect DtxR activity as dramatically as did substitutions of the amino acids involved in coordination of the anion. To further assess the relative requirements for the various residues involved in metal and anion binding at site 1, additional multiple substitutions were constructed and analyzed (Table 5) as described above. Interestingly, the effect of the H98A/E83A double substitution in strain DH5 was no greater than the effect of either mutation alone, and in strain SG22098 the H98A/E83A variant was similar to wild-type DtxR both in repressor activity and in the level of expression of the mutant protein. In striking contrast, both the H79A/E83A and H79A/H98A double-substitution variants had no DtxR activity and were not detectable as immunoreactive proteins in Western blots. Double substitutions for residues involved in the coordination of the anion at site 1 (N130A/S126A [Table 5] and R80A/S126A and R80A/N130A [Table 4]) had no greater effect than the substitutions for each residue individually. The triple substitutions R80A/H79A/H98A and H79A/E20A/R80A completely inhibited DtxR repressor activity and had greater effects than any of the single or double substitutions within the same residues, although the triple-substitution variants were produced in amounts nearly comparable to those of the single- and double-substitution variants (Table 5). In contrast, the R80A/H98A/E83A triple substitution had no greater effect on DtxR activity than did the R80A substitution alone (compare Tables 4 and 5). This was consistent with the weak effect of the H98A/E83A double substitution and the similarity in effects of the R80A/H98A and R80A/E83A double substitutions with that of the R80A substitution alone. These results further show the requirement of anion/cation binding site 1 for DtxR activity and suggest that various combinations of alanine substitutions for the ligands affect the activity of the DtxR molecule to various degrees.